Chinese AI company MiniMax is set to release its M3 large language model, which incorporates a custom sparse attention mechanism. This new model is reported to offer significant speed enhancements, with prefilling speeds up to 9.7 times faster than previous versions. AI
IMPACT Introduces a novel sparse attention mechanism that could significantly speed up LLM inference.
RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
